Improving the performance of characteristic recognition for unknown antennas with limited data

Identifying the electromagnetic features of an unknown antenna and evaluating its electromagnetic vulnerability are crucial for enhancing the effectiveness of high-power electromagnetic attacks or defenses. Sparse measurement of the radiation pattern combined with Infinitesimal Dipole Modeling (IDM) can be an effective method for antenna modeling. IDM uses mathematical representations of infinitesimal current-carrying elements to model the behavior of the unknown antenna. The more measurement data used, the higher the correlation with the original pattern. However, given the practical limitations of the observation environment, a method of obtaining high correlation with minimal data is necessary. In this paper, a method is presented to probabilistically increase the recognition performance of antenna characteristics, even when IDM is applied with limited data. This is achieved by utilizing general electromagnetic properties of antennas and data augmentation. A simulation and measurement were conducted using a rigid horn antenna. The results showed that the same correlation performance can be achieved with IDM enhanced by data augmentation, compared to traditional IDM, even when using fewer measurement points.